GitHub radar
Protect your shipped app from AI reverse-engineering agents
BLA plants decoys and artificial complexity inside your already-built app — an AI agent trying to reverse-engineer it gets stuck and burns resources, while real users notice nothing.
It's a command-line tool that injects deceptive data and extra complexity into an already-built app to confuse AI agents doing reverse engineering (code analysis).
When it helps
Useful if you ship a private app or service — like a chatbot — and want to make it harder for AI to analyze. Skip it if you have no finished build or nothing worth hiding.
Pros
- Runs on a regular laptop, no GPU needed
- Doesn't change how the app behaves for real users
- Covers many build formats — from JS to models and binaries
- No need to rebuild your project — works on an existing build
Cons
- Doesn't guarantee protection against every kind of reverse engineering
- Some file formats are claimed as supported but not yet tested by the developer
- Only available as a command-line tool for now
How to set it up — step by step
- 1Make sure you already have a built project and Node.js installed — no GPU needed.
- 2Open your AI agent (Claude Code, Codex, etc.) in your project's folder.
- 3Send the agent the text from the block below.
- 4If the tool asks for access beyond the current folder, don't allow it without checking why.
- 5Check the report on what changed in the build after running `npx bla-agents harden .`.
- 6Run the app and confirm it still behaves the same for users.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up BLA from https://github.com/qybaihe/bla — follow the README (npx bla-agents harden . on an already-built project), show me the report of what changed in the build, and ask me for permission if the tool requests access beyond the current folder.
Runs on a regular laptop — just Node.js, no GPU needed.
Open on GitHub▌ More finds